One camera, subject in front of a green screen.

[Robert McGee]

I saw a video of how they converted the first few minutes of Harry Potter and the Half-Blood Prince into 3D so I was wondering, if any of you have seen that here is my question.

If I am shooting my subject in front of a green screen with just one camera rolling, will it be much simpler to do a 3D conversion in post?

 

[GPSchnyder]

Yeah, as you can position the actor in 3D. If you have 3d in the background and render it from both eyes, replace the background matching the "eye" you render out of ie After Effects it will you a look similar to Alice in Wonderland. Flat people in 3D. Will work if the whole movie is recorded that way.

 

[Pedro Guimaraes]

actually what alice did was model the subject in 3D and wrap the flat image around it. Like a texture wrap.....better but still no replacement for shooting it in 3D.

 

[Eric Lange]

What we're working with is that you shoot 3d (stereo). You build 3d substrates or partially lower res “wire frame” models of the scene/subject from the stereogram. Then we correctly map the whole stereogram onto the 3d substrates, i.e. stereo texturing not mono texturing (we do this very precisely so there is no mismatch with the geometry). What this gives you are completely life like 3d models indistinguishable from the original stereogram, but with the advantage that you can re-render it from substantially different positions, and also assemble much larger Macro models from stereograms shot from different positions and orientations. It’s nice because the natural light and complexity and depth of surfaces (stereoscopic specular surface reflections and speckly shadows) etc are all, in effect baked in. Also for real time rendering the basic paradigm (using Coherently Stereo Textured Models) is 400x more efficient for real time rendering over conventional models that need to represent the same level of complexity (definite applications for s-3d video games). It also has the accidental (industrial by product) that it can seamlessly re-arrange the parallaxes within the scene so that surface 3d textures are maintained (preserved as rich and engaging), but the macro parallax of the whole scene can be reined into very comfortable limits. This means you can shoot with a wide IOD to capture fine 3d texture but then re-arrange the parallaxes of the scene to a total comfortable range of parallax. [you wouldn’t want to render a whole movie that way (unless you need to re-sample all of the existing parallaxes into more comfortable limits), but for a sticky shot where wide IOD was used, it may be an ideal technique.] The software we built also has a stereo 3d cursor (and various tools) that allows you to completely manually (if you wish), to build very accurate and clean stereo textured models in a full stereo immersive environment (what you see is what you get). This enables you to make very efficient and clean models, not shaggy looking automated weirdness.

At the moment for VR applications and the like we are building high precision (metrologically accurate) medium format stereo digital metric cameras, but we would love to scrounge or partner with folks that have some nice, well shot high res, stereo sequences so that we can build these new types of models and demonstrate their usefulness in cinematic applications, as well as looking for friendly beta testers to kick the software (hard).

It really is exactly where s-3d meets CG on the street corner.

Nothing is lost from the original stereogram, S-3d bites CG on the arse and CG makes “breakfast in Bed’ for S-3d.

We are a small startup and our US patent for the technique was granted just this past January, (after many years of toil etc). But looking to do a bit of useful partnering as high end software development and camera engineering is expensive enough (as it is) without having to pay for the whole shooting match! Would be nice to hook up with shooters that understand what a rich and well prepared stereogram really is, and folks that are interested in pushing the frontiers of what the digital work flow can bring to s-3d and beyond. [We are based in New Mexico (that's in the US);as well as the UK.]

With all the s-3d and full 3d stuff we are all just scratching the surface.

Cheers,

Eric


P.S./Brain fart: the coherently stereo textured models or CSTM’s are backwards compatible with conventional (mono) textured models and can be exported in standard formats, so like what Robert (above) is looking to do you can build very nice clean conventional textured mapped models from stereograms and stereographic sequences using the software we built. (Stereo IS always better than mono!)

 

[Bruce Allen]

It also has the accidental (industrial by product) that it can seamlessly re-arrange the parallaxes within the scene so that surface 3d textures are maintained (preserved as rich and engaging), but the macro parallax of the whole scene can be reined into very comfortable limits.

I was talking about that in another thread! Eg a kind of HDR tone mapping, except in Z-space. Accentuate the fine details while managing out the big transitions cleverly to fit within the depth budget (as opposed to HDR tone mapping where you're trying to fit within a display gamut "luminance budget"). Except I was thinking of shooting with 3 cameras - getting fine detail from more widely spaced cameras, and overall layer positions from more closely spaced cameras (and also allowing for creation of different stereo views for small vs large screens, as well as more than two views for autostereo).

We are a small startup and our US patent for the technique was granted just this past January, (after many years of toil etc). But looking to do a bit of useful partnering as high end software development and camera engineering is expensive enough (as it is) without having to pay for the whole shooting match! Would be nice to hook up with shooters that understand what a rich and well prepared stereogram really is, and folks that are interested in pushing the frontiers of what the digital work flow can bring to s-3d and beyond. [We are based in New Mexico (that's in the US);as well as the UK.]

With all the s-3d and full 3d stuff we are all just scratching the surface.

Yep! Haha, I'm sure when everyone understands the idea you're suggesting you'll have lots of partners. Until then, I'll just have to write you a nice email quickly

Bruce Allen
www.boacinema.com

 

[Robert McGee]

This is actually going to be a straight to video release. The DVD is going to be a 2D version while the Blu-Ray is going to use the 3D Blu-Ray standard.

The movie is acutally a sequel to a movie that was compiled of nothing but 2D images done in MSpaint, the sequel is going to have 3D rendered models somewhat loosly based on that style. It's basicly a sequel to this movie: http://www.amazon.com/VoiceOver-Ins...ie=UTF8&s=digital-video&qid=1273856888&sr=8-2

The live-action subject is going to be a sock puppet.

 

[Jay Gannon]

Its hammertime:

Hi Joe.

 

[Ketch Rossi]

What we're working with is that you shoot 3d (stereo). You build 3d substrates or partially lower res “wire frame” models of the scene/subject from the stereogram. Then we correctly map the whole stereogram onto the 3d substrates, i.e. stereo texturing not mono texturing (we do this very precisely so there is no mismatch with the geometry). What this gives you are completely life like 3d models indistinguishable from the original stereogram, but with the advantage that you can re-render it from substantially different positions, and also assemble much larger Macro models from stereograms shot from different positions and orientations. It’s nice because the natural light and complexity and depth of surfaces (stereoscopic specular surface reflections and speckly shadows) etc are all, in effect baked in. Also for real time rendering the basic paradigm (using Coherently Stereo Textured Models) is 400x more efficient for real time rendering over conventional models that need to represent the same level of complexity (definite applications for s-3d video games). It also has the accidental (industrial by product) that it can seamlessly re-arrange the parallaxes within the scene so that surface 3d textures are maintained (preserved as rich and engaging), but the macro parallax of the whole scene can be reined into very comfortable limits. This means you can shoot with a wide IOD to capture fine 3d texture but then re-arrange the parallaxes of the scene to a total comfortable range of parallax. [you wouldn’t want to render a whole movie that way (unless you need to re-sample all of the existing parallaxes into more comfortable limits), but for a sticky shot where wide IOD was used, it may be an ideal technique.] The software we built also has a stereo 3d cursor (and various tools) that allows you to completely manually (if you wish), to build very accurate and clean stereo textured models in a full stereo immersive environment (what you see is what you get). This enables you to make very efficient and clean models, not shaggy looking automated weirdness.

At the moment for VR applications and the like we are building high precision (metrologically accurate) medium format stereo digital metric cameras, but we would love to scrounge or partner with folks that have some nice, well shot high res, stereo sequences so that we can build these new types of models and demonstrate their usefulness in cinematic applications, as well as looking for friendly beta testers to kick the software (hard).

It really is exactly where s-3d meets CG on the street corner.

Nothing is lost from the original stereogram, S-3d bites CG on the arse and CG makes “breakfast in Bed’ for S-3d.

We are a small startup and our US patent for the technique was granted just this past January, (after many years of toil etc). But looking to do a bit of useful partnering as high end software development and camera engineering is expensive enough (as it is) without having to pay for the whole shooting match! Would be nice to hook up with shooters that understand what a rich and well prepared stereogram really is, and folks that are interested in pushing the frontiers of what the digital work flow can bring to s-3d and beyond. [We are based in New Mexico (that's in the US);as well as the UK.]

With all the s-3d and full 3d stuff we are all just scratching the surface.

Cheers,

Eric


P.S./Brain fart: the coherently stereo textured models or CSTM’s are backwards compatible with conventional (mono) textured models and can be exported in standard formats, so like what Robert (above) is looking to do you can build very nice clean conventional textured mapped models from stereograms and stereographic sequences using the software we built. (Stereo IS always better than mono!)

Very interesting Eric, would love to hear more, email me some info.

ketchrossi (@) ketchrossihumanitas.org

 

[Robert McGee]

Its hammertime:

Hi Joe.

What do you think your doing, I got the info I need here.

 

[Charles Angus]

Yes, it will be much easier to do post-3d if your FG subject is shot on a greenscreen.

You will still have to deal with the BG plate, though.

 

[Robert McGee]

A single sock puppet, so Charles your saying it's very easy to build 3d substrates of a single subject in front of a green screen or a blue screen in a 2D to 3D conversion. I'll be using a green screen, just so you know.

After the 3D conversion, I'll order 2 video renders done and that they be sicked up properly so that I can use any 3D background I want.